The structures of two major GSTs, Na-GST-1 and Na-GST-2 from Necator americanus, have been determined using recombinant protein expressed in and secreted by Pichia pastoris. Both Na-GST-1 and Na-GST-2 crystals were obtained from solutions containing a high percentage of polyethylene glycol. The best crystals were obtained by vapor diffusion in sitting drop comprised of equal volumes of 18 mg/ml protein and precipitant solution. Precipitant solutions were comprised of 0.1 M sodium acetate pH 4.6 and 30% PEG 400 for Na-GST-1. The precipitant solution for Na-GST-2 was 18% PEG4000, 0.1125 M HEPES pH 7.55, 11.25% isopropanol, 0.01 M sodium acetate and 0.06 M sodium citrate, to yield a final pH of 7.25. The statistics for data collection and refinement are shown in Table 1.

Initial phases were obtained by molecular replacement (MR) using the program PHASER 262728293031 with a polyalanine model based on a monomer of HpolGST from the nematode H. polygyrus (pdb code 1TW9) as the search model 32. The cell constants and space groups were a = 50Å, b = 81Å, c = 201Å, P2₁2₁2₁ for Na-GST-1 and a = 58Å, b = 108Å, c = 167Å β = 90.02°, P2₁ for Na-GST-2. The orthorhombic Na-GST-1 contains a tetramer per asymmetric unit which corresponds to a Matthews' coefficient of 2.3 Å/Da and solvent content of 46%. Monoclinic Na-GST-2 contains an octamer in the asymmetric unit, corresponding to a Matthews' coefficient of 2.8 Å/Da and solvent content of 55%.

Although the data set for Na-GST-2 could be processed as orthorhombic and a molecular replacement solution could be obtained in the space group P222₁ having 4 monomers in the asymmetric unit, refinement stalled at the unreasonably high R-factor of 50% (R-free 55%). In an effort to confirm the space group, we generated a new search model from a monomer of the structure of Na-GST-2 that had been refined in the monoclinic space group. Using this search model, we repeated PHASER in the automatic mode with each of the alternative primitive orthorhombic space groups. A molecular replacement solution with reasonable packing was obtained in P222₁ having 4 monomers in the asymmetric unit, and the same rotation and translational function as with the polyalanine search model. However, despite having 2Fo-Fc maps that agreed well with the model, both the R-factors and free R-factors remained above 48% even after extensive positional refinement with CNS, and/or REFMAC, taking advantage of extensive non crystallographic symmetry (NCS). Simulated annealing with CNS in either carthesian or torsional mode at the following temperatures 2500 K, 5000 K, or even 10,000 K did not reduce the free R-factor while the R-factor did not drop below 35%. The inability to refine the model in the orthorhombic space group, suggests that the space group is indeed monoclinic P2₁. Furthermore, no other orthorhombic space groups yielded a molecular replacement solution with reasonable packing and no overlaps.

The final refined monoclinic P2₁ model of Na-GST-2 indicates that differences in subunit structures prevent the higher order symmetry, and there are differences in side chain orientation across the monomers, and main chain residues; both termini; G-site vicinity (33–45); as well as in proximity to the dimerization domain (residues 101–156, 172–183) which are capable of breaking the higher order symmetry (Fig. 1). Overall the monomers of Na-GST-2 align better than those of Na-GST-1 with r.m.s. deviation of 0.16 Å versus 0.38 Å for all main chain atoms. In both structures the largest variation is observed in the regions bordering the G-site and anchoring the dimerization domain (Fig. 1). These regions were previously observed to be the regions of highest variation in the only other known hookworm GST structure, that of HpolGST from the nematode H. polygyrus (pdb code 1TW9) 32.

The final models for both Na-GST-2 and Na-GST-1 structures were obtained following iterative cycles of model building in O 33 and structure refinement in REFMAC-5 with NCS averaging 343536. Na-GST-1 was refined to R-factor 18.9% (R-free 28.3%) at 2.4 Å, while Na-GST-2 was refined to R-factor 17.1% (R-free 21.7%) at 1.9 Å.

All the main chain atoms for all 206 amino acids for all monomers were ordered in the Na-GST-1 structures however some monomers have disordered side chain residues, most prominently in the region from amino acids 105 through 132, interestingly these residues are include some of those that break the higher order symmetry of the Na-GST-2 structure. In the Na-GST-2 structure, all main chain and side chain atoms for each monomer were visible in 2Fo-Fc omit maps calculated from the molecular replacement solutions.

The main chain and side chain stereochemistry of the refined models are excellent and 98% of main chain phi/psi angles lie within the allowed region of a Ramachandran plot. More details of the quality of the structure as well as data collection are shown in Table 1. The atomic coordinates and structure factors have been deposited in the PDB under accession numbers 2ON5 and 2ON7 for Na-GST-2 and Na-GST-1 respectively.

The final refined model of Na-GST-1 has four monomers in the asymmetric unit, while Na-GST-2 has eight monomers in the asymmetric unit. As is the case with all GSTs, both Na-GST-1 and Na-GST-2 form dimers. The fundamental/active units of both structures are homodimers with the classical GST topology of Nu class nematode specific GSTs (Fig 2). Each monomer of the homodimer is related by a 2-fold axis of symmetry. The highly conserved N-terminal glutathione binding site is embedded within an alpha-beta domain while the more variable C-terminal ligand binding site is in the helical alpha domain (Fig. 2).

The primary and tertiary structures of Na-GST-1 are more like that of HpolGST than Na-GST-2 (Fig. 3). Na-GST-1 has higher sequence similarity with HpolGST than Na-GST-2 67% versus 61% the same goes for r.m.s deviation 0.84 Å versus 0.91 Å, for all main chain atoms. However, the monomers and dimers are also quite similar to that of human GST such as represented by hematopoietic prostagladin D synthase (HsGST) 37 (Fig. 3. & Fig. 4). Na-GST-1 has lower sequence similarity with HsGST than Na-GST-2 34% versus 38% but higher r.m.s deviation for alignment of dimers is 1.592 Å versus 1.434 Å, for all main chain atoms. Interestingly the regions of highest variability are along the dimer interface, most notably from helices α4 and α5 as well as the relative orientation of helix α8 (Fig. 4). The variation in these regions leads to a considerably difference in the size of binding cavity and accessibility to said cavity of the GSTs (Fig. 4). Large portions of the loop linking helices α4 and α5 are disordered in the HpolGST structure. The structures reveal that in the Nu class nematode specific GSTs the monomers form a more open embrace around the binding cavity, as compared to human GST, resulting in wider cavities that are more accessible to larger compounds. This is inline with the role of these GSTs as the major detoxification mechanism for the hookworm parasite. Furthermore, the cavity is more open in the absence of glutathione (in the case of Na-GST-1) than with glutathione bound (in the case of Na-GST-2), suggesting a conformational variation upon glutathione binding.

Although no glutathione was added to the crystallization mixture, unambiguous density for a glutathione molecule was observed in G-site of each monomer of Na-GST-2 (Fig. 5). Apparently, Na-GST-2 usurps glutathione during the fermentation process. No such density was observed in the electron density maps for Na-GST-1, instead scattered density were visible in the substrate binding cavity (Fig. 5). Interestingly, the G-site of Na-GST-1 cannot fit glutathione as it is abrogated by Gln 50 which points directly in the binding cavity blocking glutathione binding. A rotation of Gln around the Cβ will allow sufficient room for gluthatione to bind. It is plausible that a pH change may facilitate this rotation. Na-GST-1 was crystallized at an acidic pH 4.6 whereas Na-GST-2 was crystallized at a more basic pH of 7.5. Evidently, Na-GST-1 is capable of binding glutathione and like Na-GST-2 is catalytically active at pH 6.8 as measured by glutathione conjugation with 1-chloro-2, 4-dinitrobenzene (CDNB) 3839.

As was observed in other GSTs, the conserved, catalytic Tyr (Tyr 8) stabilizes the Cys moiety of glutathione. Tyr 8 forms a hydrogen bond with the sulfur of glutathione. The formation of this hydrogen bond interaction may result in a lower pKa for the thiol in the GST-glutathione complex 4041. The main chain oxygen and nitrogen of Ile 51 form hydrogen bonds with the nitrogen and oxygen of the Cys of glutathione (Fig. 5). The side chain glutamyl residues of glutathione face the inter-domain cleft and are stabilized by hydrogen bonds with Trp 39 (Fig. 5). Trp 39 is conserved in sigma class of GSTs, typified by hematopoietic prostagladin D synthase 37 whereas Na-GST-1 like another Nu class GST HpolGST has a Phe in that position. Phe 39 is incapable of forming the same type of hydrogen bond however Lys 43 is in close enough proximity to form the same interaction as Trp 39. In addition, the glycyl residue in glutathione forms hydrogen bonds with Ser 64, while having an inter-molecular hydrogen bond from the conserved Asp 97 from across the dimer interface (Fig. 5). Our Na-GST-2 structure gives the first view of glutathione binding in a Nu class GST 32.

Ligand binding or H-sites structures vary across classes of GSTs because the C-terminal H-sites are largely responsible for the varying substrate specificities of the GSTs. As was observed in the structure of HpolGST, the H-site of Nu class GSTs as represented by Na-GST-2 forms a long, deep cleft (Fig. 6). This cleft is formed by the interaction between hydrophobic residues Gly 13, Ala/Leu 14, Leu/Phe 65 from the α3 domain with the residues Tyr 95, Phe/Tyr 106 and Phe 206 from the α/β domain. There is an additional stabilizing salt bridge from Glu 162 to Arg 201 (Fig 6). In the Na-GST-2 structure there are some ethylene glycol molecules from the cryoprotectant solution lining the surface of the H-site (Fig. 6). The significance of the ethylene glycol molecules remains unclear until we probe the H-site with a suitable ligand. The primary and tertiary structures of the H-sites of the Nu class GSTs overlay quite well (Fig. 3). Na-GST-1 and Na-GST-2 as was observed for HpolGST 32 exhibit H-sites or ligand binding sites that are larger and more receptive to longer, larger inhibitors which cannot fit in the smaller H-site of mammalian GST (Fig. 7). Overlaying of the cavities reveals the considerable reduction in the active site size between sigma class and nu class (Fig. 7). This, in addition to the more open and accessible binding cavity (Fig. 4) suggests that it is possible for the GSTs to play the role of the major detoxifying system in the hookworm parasite.

We examined if conventional anthelmintics were capable of competitively inhibiting the hookworm GSTs by measuring the conjugation of CDNB with glutathione. We then compared our activity data to that of model roundworm Ascaris suum (As-GST-1) which used the same assay and conditions as we did 42. Like hookworms, tapeworms are treated with albendazole or mebendazole. Our results show that albendazole has approximately 5 times lower IC50 for Na-GST-1 and 7 times lower IC50 for Na-GST-2 than As-GST-1, Table 2. The crystal structure of As-GST-1 has not been solved but comparing the primary sequence of As-GST-1 with the Nu class GSTs reveals a high sequence similarity even in the C-terminus alpha domain (Fig. 3a), suggesting similar substrate specificity and ligand binding. However, As-GST-1 some large residues which may obscure the H-site (Lys 183, Lys 203 and Trp 167), accounting for the higher IC50.

The IC50 for both albendazole and chlorotriphenyltin were higher for Na-GST-1 than Na-GST-2 (Table 2). Chlorotriphenyltin was the most effective inhibitor of Na-GST-1, Na-GST-2 as well as As-GST-1. This is not surprising based on the relatively small size of chlorotriphenyltin compared to albendazole. Chlorotriphenyltin successfully competes with CDNB for glutathione conjugation as it is more likely to have access to the H-site, than albendazole, however it is doubtful that chlorotriphenyltin is specific for any class of GSTs. In order to clarify the mode of drug binding, we are currently attempting to co-crystallize both hookworm GSTs with albendazole or chlorotriphenyltin.

Neither Na-GST-1 and Na-GST-2 are effectively inhibited by praziquantel, however, praziquantel is a broad spectrum anthelmintic and is the leading drug for the treatment of schistosomiasis. The structure of the SjGST from Schistosoma japonica in complex with praziquantel has been solved and in this structure, praziquantel appears to bind in the G-site 43. It was also observed that praziquantel does not competitively inhibit recombinant SjGST 44, suggesting that SjGST is not the target of praziquantel as was speculated by McTigue and colleagues 43. In fact based on our data we may conclude that neither Na-GST-1 nor Na-GST-2 is targeted by praziquantel.

A cDNA library of of N. americanus infective larvae (L3) was constructed as previously described 4546, cDNA encoding Na-GST-2 and Na-GST-1 were isolated by immunoscreening of N. americanus L3 cDNA expression library using antiserum against Ac-GST-1 from A. caninium followed by cDNA cloning and sequencing 45. The entire coding sequence was PCR amplified from the first strand cDNA of adult N. americanus with gene-specific primers. The PCR products were sub-cloned into the Pichia expression vector pPICZαA (Invitrogen) via the XhoI and XbaI sites. The correct insert and right reading frame were confirmed by double strand sequencing of recombinant plasmid using flanking vector primer: α-factor and 3' AOX1. Fermentation and large scale purification were carried out according to the protocols described for Na-ASP-2 47 and elsewhere (Goud et al. to be published).

All crystals were grown at 22°C by vapor diffusion in sitting drops. Both Na-GST-1 and Na-GST-2 was concentrated to 18 mg/mL in 100 mM Tris HCL pH 8.0, prior to screening for crystallization conditions. Initial crystallization screens were performed using the following commercial screens from Nextal Biotechnologies (Qiagen) Classics, PEG and Cryo. Single chunk-like Na-GST-1 crystals were obtained from 0.1 M sodium acetate pH 4.6 and 30% PEG 400 (or 25% PEG MME 500). Initial Na-GST-2 crystals were flat, thin stacked plates which were smaller than 0.03 mm on the smallest face. Following further optimization using the Nextal OptiSalt pre-filled screens, thicker rods were obtained in 18% PEG 4000, 0.1125 M HEPES pH 7.55, 11.25% isopropanol, 0.01 M sodium acetate pH 4.6, 0.06 M sodium citrate. Na-GST-1 crystals were cryo-protected by adding 15% trehalose directly to the drop. Na-GST-2 crystals were transferred to a cryo-protecting solution comprised of the precipitants and 20% ethylene glycol prior to flash cooling. Crystals were flash cooled in a stream of N2 (X-stream 2000 low temperature system, RigakuMSC) prior to data collection. The X-ray system consisted of RuH3R rotating anode generator (RigakuMSC) operating at 50 kV and 100 mA, with Osmic Micromax optics and an R-axis IV++ image plate detector (RigakuMSC). Data were collected from a single crystal using a crystal-to-detector distance of 200 mm and exposure times of 20 minutes for 0.5° oscillations.

All X-ray data sets were collected using Crystal Clear (d*trek) package 48. Data was processed using MosFLM 4950. The structures were solved by molecular replacement with PHASER 262728293031. This was followed by iterative cycles of manual model building with the program O 33 and structure refinement with REFMAC5 3536 using a maximum likelihood refinement procedure with Engh & Huber Geometric parameters and free-R 51. Data and refined model statistics are shown in Table 1. The superposition of models and r.m.s. deviations were calculated using the program Swiss-PdbViewer Version 3.7 5253. Unless otherwise noted, figures were generated using pyMOL 54.

GST activity and drug inhibition was determined in a standard assay measuring the conjugation of 1-chloro-2, 4-dinitrobenzene (CDNB) with glutathione 3839. The assay mixture contained in a total of 1 mL, 1 mM CDND, 1 mM glutathione and 0.1 M potassium phosphate buffer, pH 6.5. To determine GST activity, the rate of increase of absorbance at 340 nm, were taken with a NanoDrop ND-1000 Spectrophotometer every thirty seconds for 3 minutes upon initiation of the reaction by the addition of varying concentrations of each enzyme.

Drug inhibition was measured using a fixed concentration of enzyme (1 μM) and using concentration, from 1 nM to 20 mM, of the following compounds 1) albendazole (MP Biomedicals), 2) praziquantel (Alexis) and 3) chlorotriphenyltin (Acros Organics) to inhibit in the glutathione CNDB conjugation reaction. Water insoluble compounds were dissolved in 100% ethanol. Absorbance readings, at 340 nm, were taken with a NanoDrop ND-1000 Spectrophotometer every thirty seconds for 3 minutes upon initiation of the reaction by the addition of 1 μM enzyme. The 50% enzymatic inhibitory concentrations (IC50) were determined from direct plots of percent activity against inhibitor concentration from data points generated from at least 4 repetitions of each time point and concentration.